The Involuntary Entrepreneur

It’s 2:47 on a Sunday afternoon and I’m trying to answer a simple COO question: what are my competitors building?

Allocortech is in the drone supply chain in the US. The drone defense industrial base is, generously, larval. Half the “competitors” are LinkedIn ghosts running on Air Force SBIR fumes. The other half are press-release companies whose product pages were last updated when the press release went out. The other-other-half are hobbyist shops that would really like to grow up from being Pinnochio to being a “real boy”. Even serious players have products pages that are, shall we say… aspirational? Everyone is building, nobody is shaped like a serious manufacturer yet, and the analyst coverage is one Substack and a guy on Twitter who used to work at Anduril (in facilities, probably).

So I’ve got seventeen tabs open, three half-written notes, a competitive map that looks like a Family Circus dotted-line cartoon, and the increasing conviction that the question I’m supposed to be answering is not actually answerable today.

This is not an atypical Sunday afternoon malfunction. And it’s the reason I keep starting companies.

I have a stupidity detector. It fires whether I want it to or not. I cannot turn it off, I cannot pretend the alarm isn’t going, I cannot finish anything else while it’s screaming. This is a wiring problem, not a flex. The founder essays you’ve read mostly skip this part or lie about it, because it isn’t actually flattering.

The detector has two registers. The first is moral-aesthetic: this is fucking stupid. Someone has built or sustained a thing that should not exist in this form, and the detector does not respect their feelings about it. The second is friction-tedium: there’s surely a better way? Some workflow taxes the human who has to do it, and the tax is so obviously avoidable that watching anyone pay it makes me literally nauseous.

Same internal sensation. Different triggers. Different kinds of companies fall out of them, which is something I didn’t understand until I started writing this down.

My involuntary detector shows up in two specific market shapes, and they’re almost opposites.

Larval markets. Nothing has been built. In my COO example (today), the boring infrastructure for answering basic questions doesn’t exist. Every operator is rebuilding the same maps in private, badly, and most of them give up halfway and just guess. The detector fires on absence. The drone space right now is larval. The competitive landscape my COO question wants is genuinely not in any database. I am not failing to find it. It is not there to find.

Fossilized markets. Something was built, thirty years ago, and the industry has Stockholm-syndromed itself into believing this is just how it is. The “solution” exists, it’s terrible, and everyone using it has acclimated. The detector fires on normalized badness. This one is harder to trust, because you have to disbelieve an entire industry’s collective shrug. Everyone using DOORS knows DOORS is bad. They’ve made peace with it. They have spreadsheets they email each other called Requirements_Final_v17_REAL_FINAL.xlsx. And when you say “this is fucking stupid,” the response is not “you’re right” but “well, that’s just what enterprise software is like.” (This is why I built RTMify.io, by the way.)

Both market shapes fire the same alarm in my brain. Both produce the same internal invective-laden sentence in my head, even though they are structurally inverse problems.

One more example.

November 2025: Dexcom got FDA clearance for a product called Smart Basal. It’s the first CGM-integrated basal insulin dosing optimizer for Type 2 diabetes. The press release went out November 19. Peter Simpson, SVP of Innovation, landed the quote. The release describes it as “designed to remove barriers to insulin initiation.” The 510(k) filing describes a clean architecture: HCP-set parameters, server-side titration service, daily dose recommendations driven by all the CGM data instead of one fasting value, dose reduction after lows.

Smart Basal? I conceived it, led the team that built it, pushed it into IDE and feasibility trial and slogged on for years trying to push it through Dexcom’s product development pipeline beast. Five-ish years of my life.

My name is not in the press release. It is not in the 510(k). This is standard for regulated-systems work inside a public company, and unsurprising. I’m not gretzing about that, the point is my involuntary detector.

Basal insulin titration was a fossilized market. The standard of care for adjusting long-acting insulin doses was, for thirty years, essentially “rocks and sticks”: paper algorithms keyed off a single fasting blood glucose number. People had CGMs in their pockets streaming a glucose value every five minutes, 288 readings a day, but no one was using it for titration.

The detector fired so hard it left a mark. Why are we titrating off a single fasting value when we have continuous data? Same shape, exactly, as: Why is requirements traceability a forty-thousand-dollar single-seat enterprise license when it could be a flimsy little local program?

That second question is RTMify. Fossilized market, current bet, same detector, this time I own the outcome. (Mostly. The detector is firing on three other things while I write this. Pour one out.)

So the detector fires. So what?

Being blunt, I am a worse employee than I should be, for one. The detector does not respect chain of command. It fires inside other people’s companies, on other people’s workflows, on systems that were chosen by committees I was not on for probably-valid reasons I do not have access to. I have spent twenty-five years learning to swallow the alarm bells and stuff down the nausea of seeing stupid stuff normalized. I am not great at swallowing it. (I have, to be fair, gotten better. The trick is to write the alarm down somewhere instead of acting on it. The notes file is enormous.)

Often, I cannot enjoy products that work fine for everyone else. My wife asks if I want to use some app and I have to actively suppress a forty-five-second monologue about why the navigation hierarchy is structurally wrong. She does not want the monologue. Nobody wants the monologue. I myself do not even want the monologue. My involuntary detector does not care.

I have started companies that, by any reasonable assessment, do not make sense to start. Although I didn’t start it, the company I’m at right now is a bet based on my detector. Drone right now is larval. Nobody knows what wins. The companies that are going to matter in five years probably don’t have names yet. Allocortech is a bet that the detector is right about a thing the market hasn’t priced. It might not be right. The detector does not, as far as I can tell, calibrate.

And, the part specific to right this moment: when the detector fires in multiple market shapes at once, the involuntary part becomes a real liability. The Allocortech competitive question doesn’t get answered, because halfway through trying to answer it I have started sketching the tool that would have answered it. (I will not build this tool. Probablynot. Maybenot. Ok, fine, I’m building it.) The Sunday afternoon ends with seventeen tabs, no competitive analysis, an architectural sketch I do not need, and a slightly clearer picture of why I am tired than I had at noon.

The Smart Basal thing is the cleanest possible illustration of the tax. You can be the person who conceived a thing and led the team that built it through feasibility, and six years later the regulatory filing will not carry your name in any way that matters, and the press release will quote someone who joined the company after the algorithm was already in trial. The detector will still fire on the next thing. It does not reward you. It does not respect your calendar. It just buzzes in your head, inexplicably.

I am not telling you this to be dramatic about Dexcom. Smart Basal is pretty cool. Patients will get better dosing. The world is incrementally less stupid because the detector fired in 2019. That’s the actual point of doing the work, and nobody can take it. (Not even the press release.)

But the credit-shaped hole is part of the shape of the wound, and pretending otherwise is dishonest. Most of the firings produce nothing. Some produce companies someone else gets credit for. A few produce companies you actually own. The detector does not know the difference. It fires whether you’ll be credited or not, whether the timing is right or not, whether you have the bandwidth or not, whether the market is larval or fossilized or one of the boring middle states where the detector should — for the love of God — stay quiet.

Yet it does not stay quiet.

So I am, by my best understanding, an entrepreneur involuntarily. Not because I love risk. Not because I’m bold. Because the alarm is going, and I cannot work in a building where the alarm is going.

Which is, on a Sunday afternoon in April of 2026, why I have seventeen tabs open and no competitive analysis and the beginnings of a product I am not going to build (ok that’s a lie) and a clearer picture of why I’m tired than I had this morning.

The detector is firing again.

I should probably go answer it.